8-48
Video Decoder Module
Frame I3 of the new sequence gets decoded but has to be kept in the
frame store until the first field of frame P6 is decoded. So, the Video
Decoder repeats frame P2.
In case 2, the new sequence does not arrive until some time after the
sequence end code, so frame P2 has to be repeated several times. Case
3 shows a bitstream with single pictures in the sequences and intentional
delays between sequence ends and starts. The single pictures are
continuously repeated between sequences.
Case 4 in
Figure 8.10
shows the situation where the host has ordered a
continuous skip of B pictures and skips three of them immediately before
a sequence end code.
Since there is likely to be a delay between a sequence end code and the
next sequence start, it is practical to display the last anchor picture at the
sequence end instead of waiting for the first anchor picture in the new
sequence. In Rip Forward Mode, the decoder stalls at the sequence end
code until the Rip Forward Single Step Command bit in Register 238
(
page 4-53
) is set.
As described, the last anchor picture in a sequence is displayed after the
sequence end code is detected and is treated as a still picture until the
next sequence start code. If the host sets the Ignore Sequence End bit
in Register 239 (
page 4-55
), the last picture in the current sequence is
not displayed until after the next sequence start code. This feature is
useful when the delay between sequences is short and adding the extra
display time could interfere with the synchronization of video and audio
processing.
8.6 Error Handling and Concealment
The L64105 can detect a variety of errors in the bitstream. The decoder
tries to conceal any errors found. This is usually done with the help of
concealment motion vectors if they are present in the bitstream.
Concealing errors helps minimize their effects and helps the decoder
resynchronize to the bitstream as soon as possible.